How AI will change chip design

The end of Moore’s law is approaching. Engineers and designers can only do so much to miniaturize transistors and insert as many of them as possible into chips. So they’re turning to other approaches to chip design, incorporating technologies like AI into the process.

Samsung, for example, is adding artificial intelligence to its memory chips to enable in-memory processing, thereby saving energy and accelerating machine learning. Speaking of speed, Google’s TPU V4 AI chip has doubled its processing power compared to that of its previous version.

But AI has even more promise and potential for the semiconductor industry. To better understand how AI is set to revolutionize chip design, we spoke with Heather Gorr, senior product manager for the MathWorks MATLAB platform.

How is AI currently being used to design the next generation of chips?

Heather Gorr: Artificial intelligence is such an important technology because it is involved in most parts of the cycle, including the design and manufacturing process. There are many important applications here, even in general process engineering where we want to optimize things. I think flaw detection is important at all stages of the process, especially in manufacturing. But also thinking about the future in the design process, [AI now plays a significant role] when you are designing the light, the sensors and all the different components. There’s a lot of anomaly detection and fault mitigation that you really want to consider.

Heather GorrMathWorks

So, thinking about the logistic modeling you see in any industry, there is always a planned downtime that you want to mitigate; but you also end up having unplanned downtime. So, looking back at historical data from when you had those moments when it perhaps took a little longer than expected to produce something, you can take a look at all that data and use AI to try and identify the cause. next or to see something that could also come up in the processing and design phases. We often think of AI as a predictive tool or as a robot doing something, but many times you get a lot of information from data through AI.

What are the benefits of using AI for chip design?

Gorr: Historically, we have seen many models based on physics, which is a very intense process. We want to build a reduced-order model, where instead of solving such an expensive and computationally extensive model, we can do something a little cheaper. You could create a surrogate model, so to speak, of that physics-based model, use the data, and then run the parameter scans, optimizations, Monte Carlo simulations using the surrogate model. This takes much less computational time than solving physics-based equations directly. So, we’re seeing this benefit in many ways, including the efficiency and economics that are the results of rapidly iterating experiments and simulations that will really help in the design.

So is it like having a digital twin in a way?

Gorr: Exactly. This is pretty much what people are doing, where you have the physical system model and the experimental data. So together you have this other model that you can tweak and fine-tune and try out different parameters and experiments that allow you to blow away all those different situations and ultimately get a better design.

So, will it be more efficient and, as you said, cheaper?

Gorr: Yes sure. Especially in the experimentation and design stages, where different things are tried. Obviously this will produce significant cost savings if you are actually manufacturing and producing [the chips]. You want to simulate, test, experiment as much as possible without doing something using actual process engineering.

We talked about the benefits. How about the disadvantages?

Gorr: the [AI-based experimental models] tends not to be as accurate as physics-based models. Of course, that’s why you run a lot of simulations and parameter scans. But that’s also the benefit of having that digital twin, where you can keep that in mind – it won’t be as accurate as that precise model we’ve developed over the years.

Both the design and manufacturing of the chips require intensive use of the system; you have to consider every little part. And this can be really challenging. It’s a case where you might have models to predict something and different parts of it, but you still need to bring it all together.

Another thing to think about is that you need the data to build the models. You have to incorporate data from all kinds of different sensors and different types of teams, and that way the challenge increases.

How can engineers use AI to better prepare and extract information from sensor hardware or data?

Gorr: We always think about using AI to predict something or carry out robotic activities, but you can use AI to work out patterns and spot things you may not have noticed before on your own. People will use AI when they have high-frequency data from many different sensors, and many times it’s useful to explore the frequency domain and things like data synchronization or resampling. These can be really challenging if you’re not sure where to start.

One of the things I would say is, use the tools available. There is a large community of people working on these things and you can find many examples [of applications and techniques] on GitHub or MATLAB Central, where people have shared beautiful examples, even small apps they have created. I think a lot of us are buried in data and just aren’t sure what to do with it, so definitely take advantage of what’s already available in the community. You can explore and see what makes sense to you and bring that balance between domain knowledge and the information you get from tools and AI.

What engineers and designers should consider whand use AI for chip design?

Gorr: Think about the problems you are trying to solve or the information you might be hoping to find and try to be clear about it. Consider all the different components and document and test each of these different parts. Consider everyone involved, explain them, and distribute them in a way that is sensible to the entire team.

How do you think AI will affect the work of chip designers?

Gorr: It will free up a lot of human capital for more advanced tasks. We can use AI to reduce waste, to optimize materials, to optimize design, but then you still have that human being involved whenever it comes to making decisions. I think it’s a great example of people and technology working hand in hand. It’s also an industry where everyone involved, even in the production area, needs to have some level of understanding of what’s going on, so this is a great industry for advancing AI thanks to how we test things and to how we think about it before putting them on the chip.

How do you imagine the future of AI and chip design?

Gorr: It really depends on that human element: involving people in the process and having that interpretable model. We can do a lot with the math minutiae of modeling, but it depends on how people use it, how everyone in the process is understanding it and applying it. Communication and involvement of people of all skill levels in the process will be really important. We will see less of those super accurate predictions and greater transparency of information, sharing and that digital twin, not just using AI, but also using our human knowledge and all the work that many people have done over the years.

From your Articles site

Related Articles Around the Web